SQL “IN” CLAUSE Performance issue

Question

I have a simple SQL statement:

select * from Employee 
where LEFT(REPLACE(EmployeeName,'.',''),4) IN ('ABCE', 'BCDS', 'EDSC', 'XDSD', 'EFSE')

And I am trying to do this in Linq:

Employees.Where(x => new[] { "ABCE", "BCDS", "EDSC", "XDSD", "EFSE" }.Contains((x.EmployeeName.Replace(".", "").Substring(0, 4))));

but the SQL statement it generates is not efficient. The EmployeeName is updated everytime before comparing to the strings instead of just once:

SELECT 
[Extent1].[EmployeeID] AS [EmployeeID], 
[Extent1].[EmployeeName] AS [EmployeeName], 
[Extent1].[EmployeeTypeID] AS [EmployeeTypeID], 
[Extent1].[Active] AS [Active]
FROM [dbo].[Employee] AS [Extent1]
WHERE (N'ABCE' = (SUBSTRING(REPLACE([Extent1].[EmployeeName], N'.', N''), 0 + 1, 4))) 
OR (N'BCDS' = (SUBSTRING(REPLACE([Extent1].[EmployeeName], N'.', N''), 0 + 1, 4))) 
OR (N'EDsC' = (SUBSTRING(REPLACE([Extent1].[EmployeeName], N'.', N''), 0 + 1, 4))) 
OR (N'XDSs' = (SUBSTRING(REPLACE([Extent1].[EmployeeName], N'.', N''), 0 + 1, 4))) 
OR (N'EFSE' = (SUBSTRING(REPLACE([Extent1].[EmployeeName], N'.', N''), 0 + 1, 4)))

How can I make the generated SQL look more like the original sql statement? Thanks

Answer 1

I only tested it with linqpad but i think this will generate a more efficient query.

var result = from record in Employee
             let name = record.EmployeeName.Replace( ".", "" ).Substring( 0, 4 )
             where new[] { "ABCE", "BCDS", [...] }.Contains( name )
             select record;

Answer 2

This is strange, but I rewrote your query and generated sql statement was

WHERE (SUBSTRING(REPLACE([Extent1].[EmployeeName], N'.', N''), 0 + 1, 4) IN (N'ABCE', N'BCDS', N'EDSC', N'XDSD', N'EFSE')) AND (SUBSTRING(REPLACE([Extent1].[Name], N'.', N'')

I guess that depends on used database and framework (in my case SQL Server 2008 and EF).

If you are using EF, you can force EF to use your sql statement

context.Set<Employee>().SqlQuery("select * from Employee where LEFT(REPLACE(EmployeeName,'.',''),4) IN ('ABCE', 'BCDS', 'EDSC', 'XDSD', 'EFSE')");

Answer 3

I think you can't (easily, at least) control such detail of the sql statements generation process.

But, if you are looking for the best possibile query, I would give a try to the LIKE operator (by making use, in your LINQ query, of a serie of .StartsWith conditions).

LIKE can use sql indexes, which is probably not happening with SUBSTRING and REPLACE .

Answer 4

I think you are comparing a character array instead of a string. Try this

Employees.Where(x => new string[] { "ABCE", "BCDS", "EDSC", "XDSD", "EFSE" }
         .Contains((x.EmployeeName.Replace(".", "").Substring(0, 4).ToString())));